NVIDIA’s $500B AI Plan Backed by Coherent’s Texas Fab

Coherent has begun construction on an expanded manufacturing facility in Sherman, Texas, increasing production of 6-inch indium phosphide wafers, a critical component across NVIDIA’s entire AI infrastructure. The expansion addresses the growing need for optical connectivity within increasingly powerful AI systems, enabling data transmission at the speed of light between chips, servers, and data centers. NVIDIA founder and CEO Jensen Huang and Coherent CEO Jim Anderson attended the groundbreaking ceremony, signifying a strong partnership aimed at bolstering domestic semiconductor manufacturing. Huang said during a conversation with Anderson that AI is a broadly impactful general-purpose technology. This move, supported by a 50 million CHIPS Act grant and NVIDIA’s commitment to a multibillion-dollar purchase commitment alongside up to 500 billion of AI infrastructure in the U.S., aims to close the gap in domestic supply chains for these essential compound semiconductors.

Coherent’s Sherman Facility Scales 6-inch Indium Phosphide Production

This is not simply an increase in semiconductor manufacturing capacity; it represents a shift towards larger wafer sizes, directly addressing limitations in the supply of essential materials for high-speed data transmission. While silicon fabrication commonly utilizes 12-inch wafers, much of the world’s indium phosphide production remained constrained to the smaller 3- and 4-inch formats, impacting yield and component availability. Transitioning to 6-inch wafers more than quadruples the usable area, significantly reducing costs and enabling the scale required to meet the demands of the rapidly expanding AI landscape. Huang emphasized the broader implications of this investment, explaining how advancements in intelligence impact every industry.

The facility’s output is not a finished product, but rather the lasers, transceivers, and optical modules that facilitate data transfer within NVIDIA networking systems, enabling different parts of the AI infrastructure to function cohesively. Anderson explained that as AI systems grow larger and more powerful, connectivity is as important as compute, highlighting the crucial role of Sherman in building this. This increased capacity is particularly vital for systems like the NVIDIA Vera Rubin Ultra NVL576, which links hundreds of thousands of processors across vast data centers. As signaling rates increase, traditional copper wiring becomes insufficient for these distances, necessitating a shift to silicon photonics. Huang explained that light offers a more power-efficient solution at this scale. Anderson concluded, signaling a renewed focus on U.S.-based production of essential AI components.

NVIDIA-Coherent Partnership Drives U.S. Optical Semiconductor Investment

The demand for optical connectivity supporting artificial intelligence is reshaping semiconductor manufacturing in the United States, with a concentration of investment now focused on Sherman, Texas. This is not simply about increasing capacity; it’s a move towards a larger wafer size that improves manufacturing efficiency. This partnership extends beyond a simple supplier-customer relationship; NVIDIA is investing 2 billion in Coherent to bolster research and development, future capacity, and U.S.-based manufacturing, alongside a multibillion-dollar purchase commitment for advanced laser and optical networking products. The need for these advanced wafers stems from the limitations of traditional copper interconnects in large-scale AI systems. The expanded Sherman facility is poised to become a central hub for building the infrastructure and supporting more than 550 direct jobs, and thousands of jobs, both direct and indirect.

Silicon Photonics Enables NVIDIA’s NVL576 Data Center Scale

Coherent’s expansion in Sherman, Texas, alters the scale at which artificial intelligence systems can operate. This move addresses a specific bottleneck in high-performance computing: the limitations of copper interconnects when linking vast numbers of processors. The NVL576, linking 576 GPUs across eight racks, demands a solution beyond traditional metal traces. Optics offers a significant advantage in power efficiency; while requiring an initial conversion from electrical to light signals, it allows for nearly limitless distance thereafter, a crucial factor when considering the energy demands of large-scale AI. This is not a new partnership for NVIDIA and Coherent, who have collaborated for roughly two decades, but the deepened relationship, including a 2 billion investment from NVIDIA alongside a multibillion-dollar purchase commitment, signals a long-term commitment to domestic manufacturing. The significance of transitioning to 6-inch wafers cannot be overstated.

Coherent’s facility is manufacturing the lasers, transceivers, and pluggable optical modules that form the connective tissue of NVIDIA networking. The factory’s output will be integral to NVIDIA Spectrum-X Photonics and Quantum-X Photonics switches, solidifying the link between advanced manufacturing and the future of accelerated computing.

CHIPS Act Funding Supports Domestic AI Infrastructure Growth

The escalating demand for artificial intelligence is directly influencing the expansion of specialized semiconductor manufacturing within the United States, with a particular focus on the optical components crucial for data transmission. These wafers are a vital component throughout NVIDIA’s entire AI infrastructure, highlighting a strategic move to bolster domestic supply chains for these essential materials. This expansion is significantly supported by public funding, including a 50 million grant from the CHIPS Act, funded at roughly 50 billion, and an initial 17 million investment from the Texas CHIPS program and the Sherman Economic Development Corporation, demonstrating a multi-tiered commitment to re-establishing American leadership in semiconductor production.

NVIDIA’s commitment to invest up to 500 billion in U.S.-based AI infrastructure through partnerships further amplifies this momentum. Huang stated that Coherent is a vital company for the future of artificial intelligence and reindustrializing the United States. The importance of these compound semiconductors, like indium phosphide and gallium arsenide, often remains behind the scenes, yet they are fundamental to the high-speed networking and optical interconnects that underpin modern AI systems. As AI models grow in complexity, the need for rapid data transfer between processors becomes paramount.